Brian Barasa

Key Skills

Software

CVAT

OneForma

Sama

Scale AI

Toloka

Telus

V7 Labs

Internal/Proprietary Tooling

Top Subject Matter

No subject matter listed

Top Data Types

3D Sensor

Audio

Image

Text

Video

Top Label Types

Bounding Box

Classification

Data Collection

Entity Ner Classification

Object Detection

Point Key Point

Polygon

Polyline

Segmentation

Text Generation

Transcription

Translation Localization

Freelancer Overview

I am an experienced AI and data annotation specialist with over 7 years working on high-quality labeled datasets for machine learning, computer vision, and autonomous systems. My background spans 2D and 3D annotation, including bounding boxes, polygons, semantic segmentation, LiDAR point cloud annotation, and sensor fusion for domains like autonomous driving and object detection. I am skilled in using tools such as CVAT, Labelbox, V7 Darwin, and proprietary platforms, and have hands-on experience with multi-stage QA, data validation, and process optimization. I take pride in consistently delivering accurate, production-ready training data, collaborating with cross-functional teams, and enhancing AI model performance by maintaining strict quality standards and adapting to evolving project requirements.

Labeling Experience

Train AI/Annotator

Internal Proprietary Tooling3D SensorBounding Box

I worked as a 3D Data Annotator at Digital Divide Data (DDD) on a 6-month contract, contributing to AI and machine learning projects that required high-precision 3D data labeling, primarily for computer vision systems used in autonomous and advanced perception technologies. My primary responsibility involved 3D annotation and labeling of LiDAR and sensor-based data, often synchronized with 2D imagery, to support model training and validation. Annotating 3D point cloud data using cuboids, polylines, and segmentation techniques, abeling objects such as vehicles, pedestrians, cyclists, road infrastructure, and environmental features, Performing 3D bounding box annotation with accurate positioning, orientation, and dimension alignment, Associating 3D annotations with corresponding 2D camera views where required, Classifying object attributes such as motion state, occlusion, truncation, and visibility and Handling complex edge cases including overlapping objects, sparse point clouds.

I worked as a 3D Data Annotator at Digital Divide Data (DDD) on a 6-month contract, contributing to AI and machine learning projects that required high-precision 3D data labeling, primarily for computer vision systems used in autonomous and advanced perception technologies. My primary responsibility involved 3D annotation and labeling of LiDAR and sensor-based data, often synchronized with 2D imagery, to support model training and validation. Annotating 3D point cloud data using cuboids, polylines, and segmentation techniques, abeling objects such as vehicles, pedestrians, cyclists, road infrastructure, and environmental features, Performing 3D bounding box annotation with accurate positioning, orientation, and dimension alignment, Associating 3D annotations with corresponding 2D camera views where required, Classifying object attributes such as motion state, occlusion, truncation, and visibility and Handling complex edge cases including overlapping objects, sparse point clouds.

2025 - 2025

Train AI/Annotator

SamaVideoBounding BoxPolygon

At Sama (formerly Samasource), I worked on a wide range of data annotation and AI training projects supporting the development of machine learning systems across computer vision, natural language processing (NLP), speech, and generative AI. My work contributed to the creation of high-quality, ethically sourced training data for global enterprise clients in multiple industries. I performed image and video annotation tasks including image classification, object detection using bounding boxes and polygons, semantic and instance segmentation, keypoint and landmark annotation, and scene-level attribute tagging. I also worked on frame-by-frame video labeling for object tracking and activity recognition, ensuring precision even in complex or occluded scenarios.

At Sama (formerly Samasource), I worked on a wide range of data annotation and AI training projects supporting the development of machine learning systems across computer vision, natural language processing (NLP), speech, and generative AI. My work contributed to the creation of high-quality, ethically sourced training data for global enterprise clients in multiple industries. I performed image and video annotation tasks including image classification, object detection using bounding boxes and polygons, semantic and instance segmentation, keypoint and landmark annotation, and scene-level attribute tagging. I also worked on frame-by-frame video labeling for object tracking and activity recognition, ensuring precision even in complex or occluded scenarios.

2021 - 2025

AI Trainer

Scale AIImageBounding BoxPolygon

I worked on a range of projects supporting the development, evaluation, and improvement of machine learning and large language models across NLP, computer vision, and multimodal domains. My scope of work covered both data preparation and quality assurance, ensuring training datasets met strict accuracy, consistency, and ethical standards. my work contributed to producing high-quality, reliable training data that improved model performance, reduced hallucinations and bias, and enhanced real-world usability of AI systems.

I worked on a range of projects supporting the development, evaluation, and improvement of machine learning and large language models across NLP, computer vision, and multimodal domains. My scope of work covered both data preparation and quality assurance, ensuring training datasets met strict accuracy, consistency, and ethical standards. my work contributed to producing high-quality, reliable training data that improved model performance, reduced hallucinations and bias, and enhanced real-world usability of AI systems.

2018 - 2021

AI TRainer/Annotator

CVATImagePolygon

In this project, I focus on tracking and annotating soccer players, their jerseys, and specific visual signs using computer vision techniques. The objective is to create a high-quality annotated dataset that can be used for tasks such as player identification, team analysis, and event recognition in soccer matches. I annotate the following elements: Players, assigning bounding boxes and consistent tracking IDs across frames. Jerseys, focusing on the jersey region to support number recognition and team classification. Signs, which include predefined visual cues such as referee signals, field markers, or advertisement boards.

In this project, I focus on tracking and annotating soccer players, their jerseys, and specific visual signs using computer vision techniques. The objective is to create a high-quality annotated dataset that can be used for tasks such as player identification, team analysis, and event recognition in soccer matches. I annotate the following elements: Players, assigning bounding boxes and consistent tracking IDs across frames. Jerseys, focusing on the jersey region to support number recognition and team classification. Signs, which include predefined visual cues such as referee signals, field markers, or advertisement boards.

2019 - 2019

Education

Work History